视觉测量中圆形标志点的全场自动识别和提取

Automatic recognition and extraction of circle marked point in full images of photogrammetry

针对工业视觉测量中复杂背景下的圆形标志点自动识别和提取问题,在利用拟合法进行标志点中心提取的传统技术基础上,提出判别标志点成像质量的形状误差,不仅可以避免其他亮斑对于标志点定位的干扰,而且可以自动剔除具有成像缺陷的标志点,并能保障和控制后续点中心定位的精度。应用于摄影测量,不仅可以达到亚像素的定位精度,而且提高了标志点识别能力和识别质量。

An automatic marked point extraction method in the whole image is proposed in this paper. Based on traditional ellipse fitting method ,this method proposes a shape error for imaging quality assessment, which not only avoids background spots＇ effect greatly, excludes shapely defective points, but also guarantees and controls spot center location accuracy. Application of this method into photogrammetry results in subpixel accuracy and improvement in marked point detection ability and quality.